texture and morphology based conductivity analysis of...

Indian Journal of Engineering & Materials Sciences

Vol. 24, August 2017, pp.261-269

Texture and morphology based conductivity analysis of fuel cell - bipolar plate

using scanning electron microscopic images

M C Pravina*, S Karthikeyan

a, B Sathyabama

b & D Synthiya Vinothini

b

aDepartment of Mechanical Engineering, Thiagarajar College of Engineering Madurai 625 015, India

bDepartment of ECE, Thiagarajar College of Engineering, Madurai 625 015, India

Received 11 May 2016; accepted 17 February 2017

The main objective of this work is to analyze strength and conductivity of the fuel cell bipolar plate using microstructural

analysis. This paper focuses on the structural characterization of fuel cell by texture and morphological analysis of its

composite bipolar plate which is the major component of the proton exchange membrane (PEM) fuel cell stack. The

composite plates were prepared with compression molding technique using activated charcoals and epoxy (araldite) at

various temperatures ranging between 170°C and 190°C with the pressure of 20 bar. Texture and morphological structures

of a composite are directly related with the strength and conductivity of the material. The internal structures of the

composite bipolar plate are acquired using scanning electron microscope (SEM) which provides microstructural information

of the specimen. In this paper, nature of bonding in SEM image is detected using texture and morphological features.

Texture features derived from Gray level co-occurrence matrix (GLCM) and Tamura is used to characterize the arrangement

of molecules and nature of bonding. Gray level co-occurrence matrix (GLCM) and Tamura based approach is used to

calculate texture features. These textural features analyze the internal textures of each samples prepared. Morphological

operations are used to find the area covered by each discontinuous region where the discontinuity affects the bonding and

thus the strength and conduction. The image processing approach analyses the internal structure of the prepared samples and

validates its flexural strength and electrical conductivity performed experimentally. The sample prepared at 190oC and

pressure 20 bar has met the target value defined by the US, Department of Energy (DOE) and has homogeneous internal

structure with less discontinuities and better bonding among the prepared samples.

Keywords: Bipolar plate, Fuel cell, Scanning electron microscope, Gray level co-occurrence matrix, Morphological analysis,

Texture analysis

The composite bipolar plate is a multifunctional

component within the proton exchange membrane

fuel cell (PEMFC) which conducts electric current

from the anode of one cell to the cathode of next cell

and these plates should possess functional properties

like mechanical strength, electrical conductivity, and

thermal conductivity1,2

. These properties directly

depend on its internal structure. Preparation and

analysis of composite bipolar plate plays an important

role in determining the conductivity of the fuel cell1,2

.

Over decades literatures have documented

preparation of bipolar plate using different materials

and processes. Antunes et al.3 states that there is a

strong relationship between the material used in the

manufacturing of bipolar plate and its final properties

moreover structural characteristics such as

morphology and size have decisive influence on its

properties.3 Artyushkova et al.

4 suggested textural

features, based on GLCM, provide measures of

homogeneity, randomness or directionality; inclusion

of morphological characteristics, such as roughness,

texture and shape parameters provide for more

inclusive description and therefore more complete

structure-to-property correlations of corrosion

behavior of carbon blacks.4 Newbury and Titchie

5

proposed that the combination of SEM imaging with

EDS X-ray microanalysis has given the materials

community one of its most powerful microstructural

characterization tool. Kundu et al.6 categorized

scheme of causes, modes and effect related to fuel cell

degradation and failure. In their work a number of

defects like cracks, surface roughness, orientation,

delamination have been observed and mechanism

have been proposed by which these features can lead

to catalyst-coated membrane degradation. Hermann et

al.1 reviewed various types of materials and proposed

for bipolar plates and critically evaluated their

physical and chemical properties. Williams et al.7 in

their work the interfacial area between the B4C and

SiC has been quantified and they showed the effect on

both densification and the resultant microstructure of

the composite. Harris et al.8 considers the

____________

*Corresponding author (E-mail: [email protected])

INDIAN J. ENG. MATER. SCI., AUGUST 2017

262

microstructural characterisation of composite

materials with relatively uniform feature size and

randomly distributed particles of various phases. Kim

et al.9 evaluated the electrical resistance of hybrid

composite plate in which top and bottom layers were

made up of carbon fiber with a graphite liner inside

and used matrix as phenolic resin to adhesively bond

both fibers and graphite. Sulong et al.10

used carbon

nano-tubes along with metal fillers and found that

mechanical properties were enhanced with increased

concentration of polypropylene matrix; hardness

increases with decreased concentration of

polypropylene matrix; electrical properties also

increases with decreased concentration of

polypropylene matrix. Adroja et al.11

evaluated the

mechanical and electrical properties of sandwich

bicomposites epoxy resin. Nedelcu et al.12

studied

microstructure and material properties of injection

molded Arboform parts using Taguchi methodology

with two levels and six input parameters. Du et al.13

focused on the influences of the microstructure on the

physical properties of the thin epoxy/CEG composites

(the thickness is 1 mm) and the optimum preparation

conditions are obtained. It is demonstrated that the

mechanical property and the impermeability of the

composites increases evidently with the resin content

changing from 4% to 30%. Considering the DOE

targets of the bipolar plates, the epoxy (30%)/EG

composite bipolar plate was considered to be the most

suitable candidate for PEMFCs. Hui et al.14

TGA

result shows that the novolac epoxy/NG composite

has excellent thermal stability and the optimum

processing conditions for preparing composite bipolar

plate are resin content about 15% by weight; molding

pressure 200 MPa; curing temperature 180°C. Lee

et al.15

proposed that the bipolar plate for automotive

fuel cells was developed with carbon fiber composite

by compression moulding due to the fact that

carbon/epoxy composite has not only high electrical

and thermal conductivities, but also high specific

stiffness and strength also it states that high-modulus

pitch based carbon fiber reinforced polymer composite

is a very promising material compared with

conventional materials. Hwang et al.16

established a

woven carbon fabric coated with graphite liner using an

epoxy resin as matrix medium here graphite composites

were fabricated by compression moulding. To achieve

desired electric properties, specimens made with

different mixing ratio, processing pressure and

temperature were tested. Sever et al.17

analysed

electrical and mechanical properties of high density

polyethylene and transmission electron microscopy was

used to observe the morphology of the nano-composite.

Kakati et al.18

prepared bipolar plate using compression

molding technique and evaluated electrical conductivity,

flexural strength.

Several types of materials are used for the bipolar

plates, e.g., metal plates, polymer-coated metal plates,

graphites, carbon–carbon composites and graphite–

polymer composites19-21

. The attractive properties of

graphite bipolar plate include excellent corrosion

resistance, high electrical conductivity, and low bulk

resistivity2. However, difficulties to machine flow

fields, low mechanical strength, and brittleness limit

the competency of graphite bipolar plate. Due to the

brittleness of graphite, the bipolar plate needs to have

several mm thicknesses, which make the fuel cell

stack voluminous and bulky. Therefore, researchers

are giving more attention towards the development of

composite bipolar plate. Metal is another competent

material for a bipolar plate, however, it is unable to

resist corrosion in fuel cells14

. As a candidate to fulfill

those requirements for fuel cell bipolar plate, carbon

composite has been developed extensively and reported

to be corrosion resistive14

with good performance in

electrical conduction and mechanical strength18

.

Even though many methods are available to

prepare composite bipolar plates, these methods are

deficient in analyzing the internal structure of the

plate to check its electrical and mechanical properties.

However various non-destructive testing methods like

visual, radiography, acoustic emission, ultrasonic

inspection are available; SEM image processing is a

suitable solution for this analysis. Vision based

analysis has been a fundamental open problem, with

industrial applications, such as automation and

production of quality products but those methods

mainly focused on the surface characteristics and it

fails to analyze the internal characteristics of the

materials. Here SEM image plays a vital role in

analyzing the internal characteristics which can

provide resolution better than 1 nm, so that it can

provide the spatial variations even in microstructures.

The spatial variations are analyzed morphologically to

find the nature of bonding, which is directly related to the

electrical and mechanical properties of the bipolar plate.

Methodology for Structural Characterization

Figure 1 shows the proposed methodology for

structural characterization of composite bipolar plate.

PRAVIN et al.: CONDUCTIVITY ANALYSIS OF FUEL CELL

In this work, activated charcoal with epoxy (araldite)

is prepared using the compression moulding process

with respect to various temperatures at constant

pressure. Flexural test are carried out to examine the

bending strength of the specimen (composite bipola

plate). Four point conductivity test is applied to obtain

its highest electrical conductivity. The microstructural

image of the specimen is obtained from SEM and it is

subjected to texture and morphological analysis.

Image interpretation is done based on

texture indices and morphological features to infer its

internal structural characterization.

Structural characterization

Characterization of a material’s internal structure is

fundamentally significant to examine its property for

Fig. 1 — Methodology for structural characterization

Tensile strength

(MPa)

Compressive

strength(MPa)

3445 1080

Service temperature Up to 190°C

Tensile strength 85 Nmm-2

Tensile modulus 10500 Nmm

Elongation at break 0.8

Flexural strength 112 Nmm

Flexural modulus 10000 Nmm

Compressive strength 90 Nmm-2

Chemical structure

CONDUCTIVITY ANALYSIS OF FUEL CELL - BIPOLAR PLATE

activated charcoal with epoxy (araldite)

is prepared using the compression moulding process

with respect to various temperatures at constant

pressure. Flexural test are carried out to examine the

composite bipolar

). Four point conductivity test is applied to obtain

The microstructural

image of the specimen is obtained from SEM and it is

subjected to texture and morphological analysis.

Image interpretation is done based on the estimated

texture indices and morphological features to infer its

Characterization of a material’s internal structure is

fundamentally significant to examine its property for

scientific understanding of the material. Internal

morphology is one of the most important factors

affecting the property of the material, thus the

functional performance of the bipolar plate composite.

The main reason for good mechanical and electrical

properties of the composites were the good dispersion

and strong interface adhesion of carbon

the entire polymer matrix. These factors var

different temperature, pressure and processing

conditions. This work mainly concentrates on the

characterization of internal structures with respect to

change in temperature. This characterization of

materials is predominantly done by means of non

destructive testing (NDT) methods for its accurate and

reliable internal information. The internal structure

can be interpreted by its textural and morphological

characteristics.

Preparation of composite bipolar plate

Activated charcoal and epoxy (araldit

used in this experiment, since the carbon

composite which is commercially available has a

good resistance towards corrosion, with less weight

yielding sensible electrical conductivity and

mechanical strength at nominal cost. Table 1 sh

the properties of activated charcoal and Table 2 shows

the properties of epoxy resin. .

prepared using the compression molding technique

where content of resin is 30 wt%.

agent is used; activated charcoal was dried at 75°C in

vacuum oven for 10 h to remove the moisture content.

Pre-heating of the resin was done in an oven for half

an hour at 60°C. Epoxy resin and activated charcoal

were mixed in a kneader for 2 h for uniformity. The

mixtures were introduced in an oven at 100°C before

Methodology for structural characterization

Table 1 — Properties of activated charcoal

Density

(gcm-3)

Thermal

expansion

Softening °C Particle size

2.58 5.4 846 50-mesh (0.297 mm)

Table 2 — Properties of epoxy resin

Up to 190°C 2

10500 Nmm-2

112 Nmm-2

10000 Nmm-2

2

BIPOLAR PLATE 263

scientific understanding of the material. Internal

morphology is one of the most important factors

affecting the property of the material, thus the

functional performance of the bipolar plate composite.

The main reason for good mechanical and electrical

properties of the composites were the good dispersion

and strong interface adhesion of carbon-epoxy over

the entire polymer matrix. These factors vary with

different temperature, pressure and processing

conditions. This work mainly concentrates on the

characterization of internal structures with respect to

change in temperature. This characterization of

materials is predominantly done by means of non-

destructive testing (NDT) methods for its accurate and

reliable internal information. The internal structure

can be interpreted by its textural and morphological

Preparation of composite bipolar plate

Activated charcoal and epoxy (araldite GY257) are

used in this experiment, since the carbon-polymer

composite which is commercially available has a

good resistance towards corrosion, with less weight

yielding sensible electrical conductivity and

mechanical strength at nominal cost. Table 1 shows

the properties of activated charcoal and Table 2 shows

Here the composite is

prepared using the compression molding technique

where content of resin is 30 wt%. Aradur 3986 curing

agent is used; activated charcoal was dried at 75°C in

vacuum oven for 10 h to remove the moisture content.

heating of the resin was done in an oven for half

an hour at 60°C. Epoxy resin and activated charcoal

for 2 h for uniformity. The

mixtures were introduced in an oven at 100°C before

Particle size Absorption

area

mesh (0.297 mm) 650 m2/g


264

pressing process to reduce porosity. The mixtures

(molds) were kept for 8 h in the compression moulding

machine at three different temperatures (170°C, 180°C

and 190°C) with constant pressure (20 bar). Finally, the

specimen �100 × 100 × 3 mm� is post heated in an

oven for 2 h. Figure 2 shows the fabricated specimen

at various temperatures. Acquisition of SEM image

NDT methods that are widely used in characterizing

micro-structures are X-ray diffraction, electron

microscopy (EM) including scanning electron

microscopy (SEM) and transmission electron

microscopy (TEM), and scanning probe microscopy

(SPM). Among these SEM owes its popularity for the

visualization of surfaces, to the diversity of types of

information that it can produce, and to the fact that

images and analytical information can readily be

combined.

The SEM was performed on Hitachi, Japan model

S-3000H instrument. The images were acquired at a

resolution about 3.5 nm in size with magnifications

(× 1000) having specimen size (max. 150 mm

diameter) and accelerating voltage (15 kV). Texture analysis of SEM image

Generally, electrical and mechanical properties of

the composites depends on its bonding structure;

highly bonded structures will have excellent

properties. The textural characteristics examines the

internal bonding nature. Highly bonded structure can

be identified by smooth textures with uniform gray

level distribution or correlated pixels. These uniform

pixels will occur only if there is no interuption in its

structure. The interuptions or discountinuities in its

uniform structure may occur due to voids, pores and

cracks. SEM image provides internal structure of

composite material which can be characterized by

extracting meaningful features like porosity, solidity

and textural descriptors from Haralick and Tamura.

The extracted texture features are given in Table 3.

From the set of available features the most compatible

texture features are selected by trial and error method

based on their usability in our application. The scope

of the selected features is to define the discontinuities

like voids, pores and cracks. In this regard the

following texture features are found to be appropriate.

The GLCM, proposed by Haralick is a tabulation

of how often different combinations of pixel intensity

Fig. 2 — Carbon-epoxy composite bipolar plates (a) bipolar plate samples, (b) microscopic image and (c) SEM image

Table 3 — Extracted texture features

Features Formula

GLCM

Energy

∑∑= =

gl glN

i

N

j

jip1 1

2),(

Contrast

∑∑= =

−gl glN

i

N

j

jipji1 1

2 ),()(

Correlation

yx

i j

yxjipij

σσ

µµ∑∑ −),()(

Homogeneity

),()(1

1

1 12

jipji

gl glN

i

N

j

∑∑= = −+

Tamura

Coarsness

∑ ∑=n

i

n

j

k

crs jipn

f ),(2

2

Contrast

( ) 4/14σµ

σ=conf

Roughness

Compactness 4π. area/ perimeter2

Porosity Pore volume / bulk volume

concrsrgh fff +=

PRAVIN et al.: CONDUCTIVITY ANALYSIS OF FUEL CELL - BIPOLAR PLATE

265

values (gray levels) occur in an image with fourteen

textural features among which energy, correlation and

homogenity remains meaningful descriptors for

analysing the bonding nature. Energy, a statistical

measure also called as uniformity, measures the

textural uniformity that is pixel pair repetitions. It

detects disorders in textures. High energy values

occur when the gray level distribution has a constant

or periodic form. The sample with small number of

pores and voids will have higher energy value.

Correlation measures linear dependency of grey levels

of neighboring pixels, with values of 0 for

uncorrelated and 1 for perfectly correlated pixels.

Homogeneity is a measure of self-similarity, i.e.,

measure of probability that pixels of each region have

the same values. Uniform structures without voids,

pores or cracks tend to have higher homogenity.

Tamura features provides a prominent solution for

human perception of textures at micro-structural level.

Features like coarseness, contrast and roughness have

high potential to distinguish different textures.

Coarseness (Fcrs) relates to distances of notable spatial

variations of grey levels, that is, implicitly, to the size

of the primitive elements (texels) forming the texture.

Contrast (Fcon) measures the way in which gray levels

vary in the image and to what extent their distribution

is biased to black or white. It calculates the local

variations present in the gray level image. High

contrast values are expected for heavy textures and

low for smooth, soft textures; uniform bonding results

in smooth textures. Roughness (Frgh) feature is given

by simply summing the coarseness and contrast

measures: Frgh=Fcrs+Fcon. Highly bonded structures

will have lower coarsness, contrast, roughness.

Compactness is a characteristics quantity by

calculating image or region’s area and perimeter, it

expresses the complex degree of shape. Using this

parameter the object can be extracted from the

background, which can be extended to extract the

discontinuities, pores or voids. Higher the bonding,

higher the compactness. Porosity describes voids

present in a sturcture. The voids between the particles

were occupied by the furnance atmosphere. The

vacuum created during combustion pushes the air out

of the carbon-epoxy particles which results in

pores.The porosity should be minimum for better

conductivity and strength.

Morphological analysis of SEM images

SEM image helps in characterizing physical

properties such as morphology, shape, size and

distribution of materials at the micro scale level. This

approach is based upon logical relations between

pixels, rather than arithmetic relations, and can extract

geometric features by choosing the suitable

structuring shape as a probe. In industrial vision

applications, morphology can be used to implement

fast object recognition, image enhancement,

segmentation, and defect inspection.

Morphological analysis remains a predominant

technique for determining discontinuous region and

the area covered by it. The specimen is said to be

homogeneous only if there is less number of

discontinuities; where these discontinuities occur

mainly due to improper bonding, voids and pores due

to the release of large amount of gases during

combustion process. To determine the discontinuity; the

morphological operation is done using a disk shaped

structuring element (fSE ) with radius R represented by

Eq. (1). Let x=0; y=0; be the position of the centre pixel

of the structuring element with radius r=R where i and j

the rows and columns, respectively.

[ ] rjrrirjyixN ≤≤−∀≤≤−∀−+−= ;)()( 22

≤<

=else

rNiff SE

0

012

... (1)

Reconstruction-based opening and closing are more

effective in estimating foreground markers

(discontinuities). Morphological opening-by-

reconstruction (Iobr) is an erosion (fero) followed by a

morphological reconstruction (fmor-rec) given by Eq. (2)

),( SEEroderecmorobr fIfI −= ... (2)

Morphological erosion is given by Eq. (4)

),( SEeroErode fimagefI =

Let C be the convolution sum of fero

∑=ji

SE IfC,

)).(( ...(3)

=

=∑

else

fCifI

SE

Erode0

1 ... (4)

Reconstruction is a morphological transformation

involving two images and a structuring element. The


266

reconstruction of I from Ierode , is denoted by fmor-rec , is

defined by the following iterative procedure:

(i) Initialize h1 to be the marker image, Ierode.

(ii) Create the structuring element fSE using Eq. (1)

(iii) Repeat:

hk+1= fdil (hk, fSE) ∩ I

until hk+1= hk

(iv) fmor-rec = hk+1

where

),( SEdildilate fimagefI =

>

=else

CifI dilate

0

01

where C is the convolution sum of fdil derived

from Eq. (3). Erosion retains significant

discontinuities and the subsequent reconstruction

restores its original shape. Accuracy of this

restoration depends on the similarity between its

shape and that of the structuring element, hence

choosing the shape of the structuring element plays

a vital role. Discontinuities are generally irregular

and has no definite shape hence disk shaped

structuring element remains an optimum solution

for determining them.

The extracted discontinuities are then numbered

and the corresponding areas occupied by them are

determined. If the area covered by the

discontinuities is less, then the sample possess

excellent electrical and mechanical properties.

Results and Discussion

Carbons with 30% epoxy were thermo

gravimetrically analyzed (TGA) to find the

degradation characteristics. The composite produces a

gradual weight loss region as shown in Fig. 3. The

weight loss is fairly constant between 0-170°C and

the initial weight loss occurs at 170°C due to

evaporation of small amount of moisture content and

there is drastic reduction in weight after 204°C. From

the figure it is evident that there is no degradation

occurs at the working temperature of the fuel cells

(70°C-100°C).

During combustion sintering of carbon-epoxy

particles occur at their point of contact. The sintered

carbon-epoxy particles try to flow and fillup the

pores and void spaces. It might be persumed that

number of discontinuities, pores and voids in the

sample were due to incomplete sindering of the

particles and inability of them to flow and fill up

the air spaces completely.

Flexural strength and electrical conductivity

Table 4 shows the values of electrical

conductivity and flexural strength at various

temperatures at 20 bar pressure. From Table 4, it is

seen that electrical conductivity and flexural

strength increases as the temperature increases. The

conductivity and flexural strength is maximum at

190oC having value of 107 S/cm and 44 MPa,

respectively. The above result also satisfies the

technical target value defined by the United States

of America Department of Energy (DOE) which is

>100 S/cm and >25 MPa for electrical conductivity

and flexural strength respectively13

.

Figure 4 shows the SEM micrograph results for

the samples at (× 1000) magnification. SEM images

for the original samples were acquired and are

represented as sample1, sample2, sample3 for the

temperatures 170oC, 180

oC and 190

oC, respectively.

These images were further subjected to texture and

morphological analysis.

Texture analysis

Table 5 shows the extracted GLCM and Tamura

features for the SEM image of each sample. The

energy value is higher for sample 3 which shows

that carbon-epoxy composite is uniformly

distributed compared to the other samples. For

bonding of materials to be uniform the contrast

feature should be low while homogeneity should be

high. Comparatively sample 3 gives better values

for both the features, proving to exhibit good

properties. The minimal value for Tamura features

like coarseness, contrast and roughness of sample 3

shows that it has less notable spatial variations than

the other samples, which is evident for uniformity in its

structure. This uniform distribution in structure yields

uniform bonding thus leads to fewer discontinuities;

hence conductivity and flexural strength of sample 3 are

comparatively better.

Sintering between carbon and epoxy particles plays a

prominent role in deciding the bonding strength, and

hence it’s electrical and mechanical characteristics.

Values of compactness and porosity for sample 3 are

found to be minimum than the other samples.


267

At temperature 190o, the sintering process was efficient

for the particles to flow and fill up the pores and voids

than at the lower temperatures. From these results it is

prominent that the nature of bonding is superior for

sample 3 (190oC).

Morphological analysis

Figures 5 and 6 show the results of morphological

analysis. Reconstruction-based opening and closing

are more effective than standard opening and closing

at removing small blemishes without affecting the

overall shapes of the objects. Regional maxima are

connected components of pixels with a constant

intensity value (discontinuities, voids, pores, etc.), and

whose external boundary pixels (non-defected areas)

all have a lower value. The regional maxima of

opening-closing by reconstruction are estimated to

obtain good foreground markers (discontinuities) and

these discontinuities are numerated and the area of

discontinuities for each sample is calculated.

Regional-maxima for samples 1 and 2 gives more

number of defects/discontinuities and hence results in

low flexural strength and conductivity with respect to

sample 3.

Calculating area of discontinuities

Figure 7 shows the discontinuity regions of SEM

image for sample 3, here the regions are numerated

and their respective area is estimated19

and given in

Table 6. The total number of discontinuities in sample

3 is 14 and the area is 5.4531e+06 (µm²) similarly the

discontinuities for sample 2 is 39 and sample 1 is 57

and their respective areas are 1.3320e+07 (µm²) and

9.8535e+06 (µm²), respectively. Which shows that

number of discontinuity regions and its area in sample

3 is much lesser when compared to other samples.

Fig. 4 — SEM micrograph (a) Sample 1: Temp. 170oC; and Pr. 20 bar, (b) Sample 2: Temp. 180oC; and Pr. 20 bar, and (c) Sample 3:

Temp. 190oC; and Pr. 20 bar

Fig. 3 — Thermo-gravimetric analysis: Temperature degradation

curve

Table 4 — Optimization of molding pressure and temperature at

30 % resin content

Pressure

(MPa)

Temperature

(°C)

Flexural strength

(MPa)

Electrical conductivity

(Scm-1)

20

170 32 94

180 39 105

190 44 107

Table 5 — Results for texture analysis

Texture features Sample 1

170oC

Sample 2

180oC

Sample 3 190oC

GLCM output

Energy 0.7905 0.8303 0.9083

Contrast 0.0413 0.0293 0.0179

Correlation 0.7573 0.7926 0.7579

Homogenity 0.9794 0.9853 0.9910

Tamura output

Coarsness 0.2476 0.2159 0.1884

Contrast 1.1182 0.4206 0.2157

Roughness 1.3658 0.6365 0.4042

Compactness 0.6298 0.4669 0.4375

Porosity 14.1298 19.9524 10.0823


268

Otsu thresholiding of histogram

Otsu method is used to automatically perform

clustering-based image threshold20

. The algorithm

assumes that the image posses bi-modal histogram

with foreground pixels and background pixels. In

other words, it divides the histogram plot based on the

threshold value into high and low bonded regions.

This method goes over all the probable threshold

values (ranges between 0 and 255) and then

determines the pixels and measures the spread of

pixel levels on either side of the threshold value. For

each probable threshold value the sum of the two

variances multiplied by their corresponding weights

(occurance of gray levels) is determined. The

threshold value with minimum intra class variance is

selected. For the sample with ample high bonding

region the threshold value will be greater increasing

the background which is the defect free regions.

Fig. 5 — Regional maxima of opening-closing by reconstruction (from left to right sample 1, sample 2 and sample 3 respectively)

Fig. 6 — Regional maxima super imposed on original image (from left to right sample 1, sample 2 and sample 3, respectively)

Fig. 7 — Number of discontinuities are numerated (sample 3)

Table 6 — Finding area of each objects

Discontinuities Number of pixels Area (µm²)

1 493 0.1880

2 646 0.2464

3 734 0.2800

4 584 0.2227

5 486 0.1854

6 1027 0.3917

7 3315 1.2644

8 267 0.1018

9 2273 0.8670

10 619 0.2361

11 1250 0.4768

12 388 0.1480

13 1043 0.3978

14 1150 0.4386

Total area= 5.4531e+06

(µm²)


269

Histogram of Otsu thresholding technique is

applied to all the samples and the plot is given in Fig.

8. The threshold values are estimated to be 63, 96 and

111 for the samples respectively. Sample 3

(Temp. 190oC and Pr. 20 bar) has higher threshold

thus highly bonded region is more when compared to

sample 1 and sample 2.

Conclusions

In this work, the internal structures of composite

bipolar plate are characterized by SEM image

Analysis and its results are validated mechanically

and electrically. The composite plates are prepared at

170oC, 180

oC and 190

oC

temperatures at pressure 20

bar. The experimental result shows that the maximum

electrical conductivity 107 S/cm obtained at 190°C,

with flexural strength of 44 MPa. The SEM image is

obtained for each sample, to analyse its internal

structure by texture and morphology. The internal

structure is characterized by features like porosity,

solidity and textural descriptors from Haralick and

Tamura. Morphological analysis is applied to

determine the discontinuous regions and the area

covered by it. The image processing based analysis

shows that the internal structure is more homogenous

with fewer discontinuities for the sample prepared at

190°C temperature, and hence it has better conductivity

and flexural strength which validates the experiments

conducted mechanically and electrically. This work can

be an optimal solution for applications that analyzes

relational conductivity and mechanical strength.

Acknowledgement This project is supported by University Grant

Commission (UGC), New Delhi (Grant No.: F.42-

897/2013SR).

References 1 Hermann A, Chaudhuri T & Spagnol P, Int J Hydrogen

Energy, 30(12) (2005) 1297-1302.

2 Yuan X Z, Wang H, Zhang J & Wilkinson D P, J New Mater

Electrochem Syst, 8(4) (2005) 257.

3 Antunes R A, De Oliveira M C, Ett G & Ett V, J Power

Sources, 196(6) (2011) 2945-2961.

4 Artyushkova K, Pylypenko S, Dowlapalli M & Atanassov P,

J Power Sources, 214 (2012) 303-313.

5 Newbury D E & Ritchie N W, J Mater Sci, 50(2) (2015) 493-

518.

6 Kundu S, Fowler M W, Simon L C & Grot S, J Power

Sources, 157(2) (2006) 650-656. Williams T, Yeomans J,

Smith P, Heaton A & Hampson C, J Mater Sci, 51(1) (2016)

353-361.

7 Harris W M & Chiu W K, J Power Sources, 282 (2015) 552-

56.

8 Kim K H, Lim J W & Kim M, Compos Struct, 98 (2013)

103-110.

9 Sulong A B, Ramli M I, Hau S L, Sahari J, Muhamad N &

Suherman H, Compos Part B: Eng, 50 (2013) 54-61.

10 Adroja, Pooja P, Koradiya S B & Parsania P H, Indian J Eng

Mater Sci, 20(6) (2013) 568-572.

11 Nedelcu, Dumitru, Plavanescu (Mazurchevici), Simona, &

Viorel Paunoiu, Indian J Eng Mater Sci, 22(5) (2015)

534-540.

12 Du C, Ming P, Hou M, Fu J, Fu Y, Luo X & Yi B, J Power

Sources, 195(16) (2010) 5312-5319.

13 Hui C, Hong-Bo L, Li Y & Jian-Xin L, Int J Hydrogen

Energy, 35(7) (2010) 3105-3109.

14 Lee H S, Kim H J, Kim S G & Ahn S H, J Mater Process

Technol, 187 (2007) 425-428.

15 Hwang I U, Yu H N, Kim S S, Do Suh J, Lee S H, Ahn B K

& Lim T W, J Power Sources, 184(1) (2008) 90-94.

16 Sever K, Tavman I H, Seki Y, Turgut A, Omastova M &

Ozdemir I, Compos Part B: Eng, 53 (2013) 226-233.

17 Kakati B K, Ghosh A & Verma A, Int J Hydrogen Energy,

38(22) (2013) 9362-9369.

18 Chinga G, Structural studies of LWC paper coating layers

using SEM and image analysis techniques (2002).

19 Otsu N, Automatica, 11(285-296) (1975) 23-27.

20 Kang S J, Kim D O, Lee J H, Lee P C, Lee M H, Lee Y

& Nam J D, J Power Sources, 195(12) (2010)

3794-3801.

Fig. 8 — Otsu thresholding based histogram plot (from left to right sample 1, sample 2 and sample 3, respectively)

texture and morphology based conductivity analysis of...

Documents